In order to compare mechanical measurements obtained with force platforms, film analysis and a kinematic arm, a sprint runner and a long distance runner were asked to run at different constant velocities ranging from 2.5 m.s-1 to 5.5 m.s-1. The external (EW) mechanical work reflected in the motion of the center of mass (CM) of the subject was computed using force platforms, film analysis and a kinematic arm. The differences observed on EW computations ranged from 12.3% +/- 2.5% (film analysis vs kinematic arm) to 1.4% +/- 1.8% (kinematic arm vs force platforms). These differences could be explained by (i) film analysis inaccuracy and (ii) vertical kinematic arm oscillations. Furthermore, the data obtained with a multisegment model confirm that the large discrepancy found in the literature on running efficiency is not mainly due to the accuracy of the force and/or displacement measurements, but to the different methods of estimation of the mechanical power produced by the muscles. It is suggested that the kinematic arm principle can be applied successfully for computation of CM displacements and CM energy changes during running.